What is transcription?
= The process of transforming audio data into a text format
Makes it easier to analyze>raw audio data
Aims to make it as representative as possible ->there will always be things that will be lost, rather creating a usable document for a specific analytic purpose
Why is transcription important?
1) Makes it easier to analyse -> Allows other researchers to use the transcript; increases utility of data beyond being used for a single study
2) Familiarisation-> Helps you, as a researcher, to become familiar with the data
What are the three types of transcription?
Phonetic (Jeffersonian)
= A complete record of what and how everything was said, including other noises.
- Nothing is corrected or changed
- Goal is to provide an accurate representation as possible
EG includes ‘up arrows’ for intonation, underlining for emphasis, insert pauses and laughter etc.
Full Verbatim (orthographic)
= A complete record of what was said, including other noises.
-Nothing is corrected or changed
-Does NOT include intonation and how things were said
Clean Verbatim (intelligent verbatim)
= A complete ‘tidied’ record of what was said, excluding unnecessary noises.
- Retains ‘core’ message but removes superfluous noises, fillers, mistakes etc.
- Aim to make transcripts less wordy and easier to read.
When should each type of transcript type be used?
1) Phonetic + Full Verbatim
-> When nuances of communication and researcher interpretation are central focus
-> eg focus critical qual data; how language is used to construct meaning, reading deeper into how people said things>what they said.
2) Clean Verbatim
-> When conversational nuances are less important.
-> Potentially suited to experiential data; less reading between the lines< taking ppl at their word, trying to capture the core essence of ppl’s words and perceptions.
What are the pros/cons with each transcript approach?
Phonetic and Full Verbatim:
Con-> Time consuming to produce
Con-> Can be challenging to read and interpret correctly (if there is lots of symbols and notations) -> Leads us to making transcripts difficult to read, easier to lose the central ideas and foci of the research through ‘messy’ language.
Pro -> Objective, includes everything, unlikely to leave any important information out as it should all be included.
Clean Verbatim
Pro -> faster to produce and easier to read
Con -> Too subjective -> up to transcriber to determine what is the core knowledge of what someone is saying, and what is superfluous. Can lead to important information being left out.
What are some tips to help with transcribing data?
1) Factor in sufficient time to transcribe
-> A one-minute audio can take around 3-10mins to transcribe
-> Focus groups>interviews take longer
2) Set up specific ‘keyboard binds’
-> Set up tabs or rewind audio in different increments of time without using the mouse. Keeps both hands on the keyboard so quicker (EG VLC media player)
3) Make sure you’re getting good audio quality
-> Quiet setting and good tech
-> Test out recording tech
4) Speech is messy, we don’t talk in sentences
-> Punctuation can change meaning!
-> Only use punctuation if it clarifies how something was said
-> Add speech/quotation marks
5) Create a set of rules and document them (for yourself and others)
-> ALWAYS check software is GDPR compliant
-> Microsoft teams Automated Transcription is!!
What is GDPR and why is it important?
General Data Protection Regulation
= A law in the EU that governs how personal info is processed by organizations.
Aims; to enhance individuals’ control over their personal data
Aims: to establish a framework for data protection compliance.
Failure to comply -> serious breach of research ethics can lead to financial penalties and even criminal prosecution
What is de-identification? (goal and why?)
The process of altering or removing all information which could lead to ppts being identified (after transcript written up).
Goal -> make ppts unidentifiable whilst maximizing fidelity of original transcripts (retain core meaning and ideas)
WHY -> keeps it ethical; kept confidential + ppts can consent to being identifiable. Otherwise, should remain hidden in any outputs from research (risks of identification in quotation in papers + sharing research in open access dataset).
What are the two types of de-identification?
1) Direct identifiers -> names, locations, numbers etc
2) Indirect identifiers -> contextual information, any info when combined with other info could lead to identification (not always clear).
How to de-identify a transcript?
1) Set of rules to be used consistently (check guidelines); helps you and others interpret what you’ve done.
2) Put altered text in square brackets [ ]
3) Give ppts pseudonyms (informative about demographic but confidential, not just P1 P2 etc)
4) Change specific info to more general (eg ‘I work at Bristol Uni’ changed to ‘I work at [a university]).
5) Reduce precision-> ‘I was born in May 2015’ changed to ‘I was born [roughly 10 years ago]
6) Substitute word entirely to a similar idea (‘I love cakes’ changes to ‘I love [biscuits]’)
7) KEEP A CHANGE LOG; to retain fidelity of ppts initial meaning. (eg a table of ‘original text’ vs ‘de identified text’ then ‘rationale’ for change)
What is a change log?
A log kept by researchers to retain fidelity of ppts initial meaning. (eg a table of ‘original text’ vs ‘de identified text’ then ‘rationale’ for change)
What is involved in qualitative coding?
-> Assigning labels to discrete units of text (ie coding phrases or words)
-> Goal is to capture the ‘essence’ or ‘important attribute’ of that portion of text (enabling us to separate, compile and organise data).
-> In later stages of analysis, codes are how we identify KEY IDEAS and their INTER-RELATIONSHIPS.
What is coding?
Breaking up the text into meaningful chunks, to make it easier to analyse later. To accurately identify what each portion of text is about (can assign same codes to multiple texts).
What are the types of practical basics used for coding? + evals
1) Physical media
-> use physical copies, pens and highlighters and scissors to identify key themes
PRO: cognitive benefits
CON: time consuming, limited to highlighter colours at hand
2) Digital media
-> Microsoft word; highlighter function and comments like on paper. Can copy and paste.
PRO: can always change comments and add more highlighter colours
CON: difficult for big data analysis with lots of transcripts
INSTEAD USE (in future career)-> QDAS (eg NVivo)
PRO: Most advanced option
CON: difficult to learn
When do you start THINKING about coding?
= Begins early on even if you are not formally coding till later…
EG -> Thoughts about coding start on day 1
Keep separate a: reflexive diary AND an analytic diary (only contain thoughts relating to analysis like emerging possible codes)
EG -> When transcribing; note down potential codes, and a log of quotes that relate to central ideas to RQ
When do you FORMALLY start to code?
-> Concurrent (ongoing) coding and interviewing; after each interview go away and transcribe and code the transcript.
This is suited to a Grounded Theory methodology -> when data collection and analysis are a continuous iterative process (analyse as you go)
OR
-> After a block of interviews start to code
OR
-> After all interviews start to code
What is the difference between descriptive and latent coding?
1) Descriptive (semantic/ manifest)
- captures the ‘surface level’ meaning of what ppts are saying
- links to an experiential qual approach (prioritizing ppts interpretation)
2) Latent (analytic/interpretive)
- capture some ‘underlying’ meaning, what’s implied considering context?
- links to critical qual approach (prioritizing Analysts interpretation)
EG “If we imagine our data three-dimensionally as an uneven blob of jelly, the semantic approach would seek to describe the surface of the jelly, its form and meaning, while the latent approach would seek to identify the features that gave it that particular form and meaning. “
= These are NOT mutually exclusive; likely to use both.
What is researcher-led coding
1) Researcher-led (aka deductive, directed)
= Code for specific topics/answers to the specific research question
-> begin with pre-defined codes drawn from existing theories/frameworks (deductive; based on these theories, this is what we expect to come up so should look for this)
-> interrogating/refining these theories/frameworks
What is data-led coding?
Data-led (aka inductive, undirected)
= Code everything that might possibly be relevant
-> develop codes from data, let them ‘emerge’
-> exploring phenomenon with fewer preconceptions, prioritizing ppts ideas
Do we only use researcher-led or data-led coding?
= These are not mutually exclusive, likely to use both
It’s about how you START coding…
-> Researcher led to begin coding with directed focus to open new ideas and unexpected codes arising from data.
-> Data-led to begin by capturing all data about data that you have little preconceptions about.
When do our philosophical frameworks help us in coding?
When we begin coding transcripts
When deciding whether codes are descriptive or interpretive (can be both)
When deciding whether we adopt a more researcher-led or data-led approach
How to refine (name) codes?
Must be concise, accurate and informative (eg ‘aims of science’ refined to ‘thematic saturation as aim’)
No duplicates (don’t want different names for the same code) eg ‘aims of science’ and ‘goals of science’ should be refined to just one of them
Ends up refining codes into themes: meaningful network of related ideas that are either implicit or explicit
What are some coding tips?
1) Don’t be afraid to merge and discard codes as you organise them into relationships
2) Continually question and refine code names
3) Coding is not counting! -> it’s about finding meaningful ideas that address our research questions.
4) Don’t discard single ideas from one ppt as it can be highly important
5) Contradictory ideas are also informative; helps us explore contextual differences
6) ‘Concept maps’ -> use pen and paper to map out concepts (to think about + understand inter-relationships)
7) Revisit raw data (look for contradictions to your conclusions, reflect on your influence on the codes, remember context)
-
Week 121
-
Week 2: Ethics + Philo43
-
Week 3: Research questions + Interviews36
-
Week 4: Transcribing + Coding Qualitative Data23
-
Week 5: Thematic Analysis29
-
Week 7: Reporting Qualitative Research34
-
Week 8: Critically Evaluating Research28
-
Week 9: Grounded Theory42
-
Week 10: What is IPA14
-
Week 10: How to conduct IPA12
